How Software is Built

Interviewers: Scott Swigart and Sean Campbell

Interviewee: Marc Frons

In this second interview with Marc Frons, CTO for the New York Times digital operations, we discuss the Times use of open source in their infrastructure. In specific, we talk about:

• WordPress and blogging at the New York Times
• New York Times investment in open source
• Licensing issues and corporate citizenship
• How they address open-source application support and evolution
• The role of content management systems
• The future of corporate investment in open source

Sean Campbell: Thanks for taking the time to chat. In our earlier conversation we talked about some of your digital media initiatives at The New York Times, and one thing you mentioned was blogging. Talk a little about your blogging infrastructure.

Marc Frons: Sure. About 99.9 percent of our blogs are on WordPress and have been for the past year or two.

Sean: What else did you guys evaluate besides WordPress — or was WordPress just an obvious choice?

Marc: We had some of our blogs on Moveable Type. At a certain point we decided to move over to WordPress. The move predates me, but I think one of the reasons was just the personal preference of the developers. WordPress has a lot of plug-ins available. I think it was about the same time The Times acquired About.com, which was also a big user of WordPress, so that might have tipped the scales.

Sean: There’s some recent news about The Times and WordPress. Tell us a little bit about what’s been happening there.

Marc: Sure. I think that over time the relationship between WordPress and The New York Times has deepened, and we’ve gotten closer to that company and the principals. We recently announced an investment in WordPress and took part in one of their second rounds of financing in WordPress and Automatic, which is the commercial end of the WordPress.org organization.

We’re very excited about that. As part of that deal, we’re working very closely with WordPress on NYTimes.com blogs and some special things for The New York Times that WordPress is going to help us with.

Some of our blogs today are not on our own infrastructure and are fairly customized. One of the things we’re going to do is have WordPress help us move our blogs to our own infrastructure, and do them the right way meaning a more standard configuration. We had customized them for various reasons, for scalability within our environment as opposed to scalability overall, so they are going to help us do that.

We are also working together to marry the New York Times taxonomy to the larger taxonomy of the blogosphere, so it’ll be easier for us to find relevant posts that have surfaced on blogs posted by WordPress.com, and to integrate those posts into our own site. We’re doing a little bit of that right now with another tool we own called Blog Runner.

WordPress will probably be helping us with some training and documentation for our newsroom, so any of our journalists who are new to blogging or really don’t understand what it’s all about can 1) learn how to use the WordPress system more easily, and 2) learn a little bit more about blogging in general and how to use these tools.

And finally, I think we’ll probably be working on some things around security for our environment, and some plug ins around that.

Our intention in all these things is to give back whatever we can to the community and keep it open source.

Sean: I’m just curious — and forgive my ignorance on this because I just don’t know — has The New York Times contributed in a financial way to an open source project prior to this, or is this the first time?

Marc: To my knowledge, this is the first time that we’ve invested in a company that started with an open source model.

Sean: So what’s the thought process behind it?

Marc: I think there were a couple of things. One is knowing the principals of the company. We’re very comfortable with them as business people and as technologists. It’s not often you get the opportunity to invest in an exciting Internet company or a growing software business at a reasonable multiple or any multiple that you can afford.

Finally, our confidence in the WordPress platform and what WordPress stands for as a company, and aligning its values with our own, was something that really attracted everyone here to this particular deal.

Sean: Talk about how you see their values aligning with yours, compared to just any open source project.

Marc: I think there’s a real commitment from the WordPress team for quality, for openness, for an exchange of ideas, but in a way that is respectful. The meetings we’ve had with them have led us to believe that this is a company that cares about quality, that cares about its customers, that has an evolving platform, and that’s tapping into the community but being smart about it.

One of the things a WordPress founder said was, “We listen to our customers very closely, but we don’t do everything they say.”

Sean: Right.

Marc: I think that’s really important for every company to think about. That really resonated with me, because I think that no matter what, you still need a filter on what you’re doing. There can be good communities and bad communities communities that are smart, and some that are smarter than others.

I think it’s up to the people leading a project, and especially an open source project, to be smart about what gets promoted into the code base what features and functionality one concentrates on and what one doesn’t. Often a special interest group can hijack a project, in a sense. That hasn’t happened at WordPress, as far as I’m concerned.

Sean: I know what you’re talking about, because we’ve always had a philosophy, just from an entrepreneurial standpoint, that to be effective you have to turn down certain opportunities. In business, you’ll be offered work that technically you could do, but that’s just not a smart direction for your company to go. I get really worried when I’m dealing with another company and I feel like their motto is “Say yes to anything.”

It seems that open source projects might be under the same pressure, where someone says that a certain direction will add thousands of users or enable a scenario they feel is critical, and someone on the project has to have the maturity to say, “Yes, that direction would benefit us, but it would cost us in other areas, and those costs outweigh this one benefit.” It sounds like you’re saying the WordPress folks really get that.

Marc: I think that’s true. And I think you should talk to the WordPress guys about that, if you haven’t already. It would be fascinating to understand their process of deciding what makes it into the core and what doesn’t.

I mean, one of the things that they said was, “Look, we have an architecture that allows you to build any kind of crazy plug in you want. But when something becomes really popular and a lot of people find it useful, it becomes part of the code base.”

Sean: Kind of like what they did with tagging recently.

Marc: Right.

Sean: They didn’t fold in the Ultimate Tag Warrior plug-in, they didn’t do every last thing they could do with tagging. They basically said, “We’ll support tagging. If you want something fancier, that’s what the plug in community is for.”

Scott Swigart: Sometimes people are skeptical when they hear about corporate sponsorship of open source. How would you respond to someone who says, “Does that mean The New York Times has become a special interest that WordPress now has to cater to, even if what the Times wants isn’t necessarily what the broader community wants?”

Marc: Well I don’t think we own enough of WordPress to really influence it that much — we’re a minority investor. Even if we owned a majority stake or the company outright, the values of the founders and the principals of the company are such that I don’t think they would let that happen where they would say, “The next release of WordPress will be totally customized for partner X or partner Y, and we don’t care about this community we built.” That would truly be shooting themselves in the foot, because their community of users is always going to be much larger than any single company. It would also be bad business for any company that acquired them.

Scott: Sean and I are both political junkies, so it sounds to me as if a politician explains a contribution by saying, “This contributor isn’t asking me to go in a different direction. He’s supporting me because he agrees with what I do.”

You’re saying the interests of The New York Times and the interests of WordPress were already in alignment, and it’s not about changing their direction.

Marc: Right. It’s not like WordPress was a turnaround situation and we said, “Gee, we’re just going to go in there and clean this place up and change their strategy. And, you know, they should be focusing on a different market, and building everything a different way.” We said, “We love what you guys are doing, and we’d love to be a part of it. Since we’re already using your product, we’d like to be closer together, both to have more input in what you’re doing and to hear more about your ideas for our business.”

Scott: Well, let me flip it around too. Let me be a crotchety New York Times shareholder who says, “Look guys. Focus on the news business. There are all these open source beauty pageants going on. Don’t get caught up in the hype. Show me how I’m going to see a return on my investment.”

Marc: We invested in a company that just happens to have an open source model. And as we’ve talked about before, I don’t think it’s accidental that that’s a reason for their success or one of the reasons for their success. But I think we primarily invested in them because we think they’re a good company with a very good sense of their market and their product and their audience, and of how they’re going to continue to grow and improve.

And I think the fact that open source is a part of that is a really good thing, but if they didn’t have a clear idea of themselves it wouldn’t matter whether they were open source or not. So when we look at The New York Times and our strategy from here on in, it’s really about treating ourselves as a multiplatform content, news, and information company.

If you think about it, we have a newspaper, we have a Web site, we have a mobile site. We have various means of publishing. We have our own internal content management system. We have blogs that are on WordPress. We may have other types of user-contributed content that might be hosted on WordPress or in a WordPress content management system. We have our content on more advanced devices, and we have an R&D group that experiments all the time, that looks around corners and asks, “What’s coming next?”

So for us, the fact that they are open source is terrific, but it is incidental to who they are in terms of our investment.

Scott: Right, WordPress is a leader in that space. Obviously, a media company that has a connection to that kind of media infrastructure, a media company working with the people who are working on key underpinnings of the blogging infrastructure, makes perfect sense.

Marc: We certainly thought that. And look, it’s not like they are the only ones out there. There are a lot of good companies doing open source blogging platforms. But I think what impressed us about WordPress is the velocity of its growth, as well as WordPress.com, which I think has come on quite strongly in 2007 and continues to grow nicely as the company really focuses on that mission of providing a robust blogging platform for all kinds of users.

Scott: Let me ask you a question in a completely different area. When it comes to licensing open source projects, there are a number of different licenses a project can use. Does licensing affect your decisions about what open source projects to use or how you use them? Does it matter whether it’s GPL v3 or v2, or something more commercial friendly like Apache? Do you look at things in-depth or is it just something that sort of floats through in the middle of the decision making process?

Marc: I would say the latter.

We don’t worry about it as much right now because we’re not producing software in that way. We were working with another company and were interested in licensing the product they’d built, and then one of the lead developers called me and said, “Look, we just have to talk to the lawyers because this includes some open source software and we have to make sure that we have all the licenses right and can reuse it in a commercial product, et cetera, et cetera.” And I said, “OK, you work that out with the lawyers.” But generally, it’s not an issue.

Sean: When you mentioned build software on top of open source, it made me think of something I’ve been wanting to get a perspective on. Google is an interesting example because it is considered an open source company, but it hasn’t really open sourced any of its significant IP.

With that in mind, do you see a time when The New York Times would potentially open source part of the effort you are building now? You’ve obviously built a high end Web portal for a very large media outlet. On top of that, The Times has a willingness to engage with the community that maybe a Google a pure for profit company doesn’t have to the same degree.

Marc: Sure. Just as Google has “do no evil,” one of the core values of The New York Times is to “enhance society.” That phrase is in our mission statement. And I think we touched on it a little bit last time. One of the things we did about six months ago was end Times Select, our pay product. Almost the entire Times Web site is now free of charge, or totally advertiser supported.

We think of that as an open model in terms of content. The next step for us is building APIs and applications that leverage our content, and then opening those APIs to the developer community at large so that others can display our content and applications in new and useful ways.

We are actively building those tools, and also thinking about additional ways to pursue that strategy. So on the Times site, we have a blog called “Open” that’s all about our open source efforts. Right now it’s largely if not exclusively written by our developers. We’re hoping to have our first New York Times developer day later this year. It would be modeled after the Google and Yahoo hack days. Before we do that, however, we’ll distribute a set of our content APIs so that developers outside The Times can create applications that combine our content with functionality they devise — and basically create open source applications using our technology and our content. So we take the effort very seriously. We think of it as part of our overall strategy of having a positive impact on society as well as improving our own Web site and our own news and information platform.

Scott: I’d also be interested in your perspective on two of the licensing models that exist in open source. On one end of the spectrum, you have Apache foundation licensing, which is very commercial friendly. You can take anything under the Apache license and build on top of it.

Marc: Right.

Scott: You can include it in your product, bundle it with commercial software, and you are not required to open source your derivative works.

At the other end of the spectrum you have the Free Software Foundation with GPL licensing. Its goal is to “infect” projects with its idea of licensing, such that if you extend GPL code, you have to open source your work.

This is in part because the Free Software Foundation takes the view that all source code should be free and should be open. You should not be able to take open source software and derive from it, essentially building on the shoulders of giants, and not give back. You also shouldn’t be able to patent software, the same way that you can’t patent a mathematical equation.

So with that in mind, what is your view on intellectual property in general? Do you have any sort of philosophical viewpoint about what’s right between those two extremes?

Marc: On the one hand, I can certainly see the FSF’s point: if you’re using open source code you shouldn’t turn around and try to make it proprietary — you should give back to the community. But I favor the Apache Foundation’s approach because I believe it actually encourages more innovation. In other words, if I know I can use open source code to create something new and am not restricted in terms of how I can profit from it, I’m more likely to use it and hence more likely to create something innovative. But if I have too closed a model, it’s likely someone will come along with an open one that works better. The marketplace is going to influence the business model.

I think individuals and companies should decide what to make open and what to keep proprietary. But open source as a movement has caught on because as an alternative it’s proving to be a faster path to innovation than closed models. The minute it stops being that, it stops being useful.

Scott: I’d also be interested in knowing how you guys get support for the open source software that you use. Obviously you’ve been able to build a highly scalable, functional Web portal, with lots of different features. Obviously you haven’t been stopped by any significant issues such as “This doesn’t seem to be working right” or “I don’t know how this is supposed to work.”

Obviously you guys have been able to work through any issues. With that in mind, do you use any paid support options from open source vendors? Or do you just go with the model of, “We’ve got really smart people and they’re able to use the resources at hand and figure stuff out?” How does that work?

Marc: We’ve used MySQL support in the past, and MySQL consultants. But you could argue that any number of ways. You could say that’s a consultant who just happens to work for MySQL. Or we could hire a really smart guy who knows MySQL really well. We chose the former because of expediency.

Sean: I am curious, though, if you have ever been confronted with an “abandonware” scenario. Obviously you are working on very stable projects in terms of Apache, WordPress, etc.

And of course you can have the equivalent of “abandonware” with closed source as well. For example, with Vista, companies were concerned that XP would no longer be supported past a certain date. In a closed source company it’s usually not that it’s abandoned, it’s just that support ends, which can have the same effect.

With that on the table, I’m curious about how you analyze an open source project to ask yourself the question, “How stable is this?” What elements of the project do you look at to say, “This project looks stable. It looks like it’s not going to disappear some day”?

Because sometimes open source projects disappear for political or organizational reasons that have nothing to do with their technical strength. I am curious how you go about that evaluation process.

Marc: I think that we have, by and large, stayed with very mainstream pieces of the open source stack. We haven’t really looked at anything too esoteric where you really have to worry about, “Is this going to go away?” or “If it goes away, I’ll never get any support and I’ll never understand this.”

We have closed source products in the house that we haven’t gotten vendor support on for five years because the vendor’s gone out of business. But we don’t change them very much, and they still work. In some cases they don’t, but it’s not that big a deal because we’ve moved on.

So I tend to not worry about it too much, because I consider the lifespan of most technologies to be fairly short.

Scott: So just to say that back in my own words, your model is to prepare for the fact that all technologies are built on somewhat shifting sand?

Marc: Correct.

Scott: So you do not plan on something being there for 10 years. You assume that you’re going to be constantly evolving your technology base; you’re going to want to rebuild using new technology. And that is just part of the way you guys do things.

Marc: That’s right. Because our main goal is to come out with great features and functionality, and the best news and information for our customers. Technology is the means to that end, not an end in itself. We will use what endures and if it doesn’t we’re not going to worry about it.

Scott: Right. And you are in the midst of that right now. You mentioned in our earlier conversation that you had developed your own architecture and language, and you are evolving off of that. And from your perspective evolution will just continue and continue and continue.

Marc: That’s right. As long as that language is serving us well in certain areas and is not a problem, we can keep that alive indefinitely. But we’re always trying to look ahead as far as possible and anticipate what our needs will be.

Scott: Are there open source projects that you currently are not using but are interesting to you? Something that perhaps you might see being used over the next year or so?

Marc: When I talk to my peers in the publishing technology space, I find that one of the things everybody complains about is content management systems. It’s so core and fundamental to what we do, and it was a space that was and still is dominated by few vendors.

Over the last couple of years, you’ve seen a lot of open source players come on. I’m not so much thinking of companies like WordPress here, although that is a kind of CMS, but I’m thinking more along the lines of Druple, Alfresco, and a few other systems.

Sean: So you’re talking about the typical competition between products like SharePoint, Alfresco, Druple, and other CMS systems?

Marc: Right. I see a lot of movement in the CMS world toward open source solutions. And I think that’s very exciting, because these sorts of reusable components and frameworks can help an entire industry build better systems much faster than it could before.

A lot of people saw CMS as a competitive advantage, and I’m not so sure it is. It’s one of those tools you use. And in the publishing industry, it’s something that is ripe for collaboration among people who have similar interests and needs, who may be competing with each other for advertisers, but there’s no discernable reason they should be competing on that level.

Scott: So are you seeing publishers actively collaborating with the open source in the CMS area?

Marc: It happens sometimes, but not in an open source way necessarily. Somebody will build a system and somebody else will say, “Hey, I’ve just spent X million dollars on this system,” and maybe it’s with vendor X and maybe we’ve done it ourselves. “Do you want to buy it? Do you want to license it?” And sometimes the answer is no and sometimes the answer is yes, depending on what’s going on.

Scott: Are you seeing these open source CMS players reaching out to publishers at all? Or is it just that the industry sees what the CMS players are doing as interesting?

Marc: I think it’s kind of the latter, because publishers are all over the map. You have a handful of large newspapers like The New York Times that have very specific needs, and then you have a whole bunch of smaller newspapers that have a whole set of other needs. And then you have magazines and blogs and everything in between.

If I’m an open source CMS provider, I have to sort of figure out where my target audience is first.

Scott: Maybe they go after the corporate intranet because they can make the case that every corporate intranet should be looking at their stuff. Whereas publishers are, like you said, a more specific niche. For example, MySQL does quite a bit of nonrecurring engineering to meet the needs of specific clients.

Marc: Correct.

Scott: These CMS vendors might find themselves in kind of the same situation, where people need certain things to be a certain way for their business.

Marc: Yes, and that’s the last thing you want to get into. I don’t think an open source solution is writing one custom piece of your program, for one particular company or individual, that doesn’t have much reuse.

Scott: That makes me think back to where we started this conversation. The New York Times is investing in WordPress, Citrix bought XenSource, Sun bought MySQL AB. I’ve seen people predicting that this trend will continue. What are your thoughts?

Marc: I think we will see more established companies investing in open source companies for a couple of reasons. The major one is that entrepreneurial software companies have proven that an open source model is a rapid path to innovation. They have also proven that they can find ways to make money from the growth of that model through services..

Sean: So how does a company like yours look at the Sun acquisition of MySQL? Does that somehow limit choice? Do you think people have concerns for what this will mean for MySQL, which so many depend on?

Marc: Well, I think there’s always a concern when a big, established company buys a smaller company. The smaller company’s agenda and what made it special is going to be subsumed by the agenda of the larger company.

And certainly, you have to look at who the acquirer is and say, “Does this company have a history of treating its acquisitions well and giving them room to breathe?” And I think Sun’s track record there has been kind of mixed.

But I think one interesting thing that has happened at Sun in the last couple of years is that it has, in its own way, embraced open source to a much larger extent. It used to be everything was Solaris or bust, and Solaris was a closed OS. But Sun has opened that up and done a lot of other things embraced Linux, embraced x86. The company is trying to be much more innovative in that way.

So I’m a little less concerned about it than I would be if a company with a history of proprietary only technology decided it was going to buy this company and then just make it part of the Borg.

Sean: Which brings us to the Microsoft offer for Yahoo! I was really surprised that mainstream media, CNN for example, picked up on the fact that Yahoo! is built up on a bunch of non-Microsoft technology.

What do you think that means when you marry those two business units? Do you have thoughts, predictions, comments on what some people might think is a difficult combination?

Marc: I think that if they are really smart about it, they’ll make the merger work by stressing interoperability and really thinking through the ad network issues. It depends on whether they are going to keep the Yahoo! identity alive, which certainly I would if I were them.

Scott: Marc, thanks for taking the time to talk.

Marc: Thank you.