Interviewee: Matt Mullenwegr
In this interview we talk with Matt Mullenweg of WordPress and Automattic. In specific, we talk about:
- Successfully making the culture shift when an open source project is acquired by a corporation
- Scaling skills and physical resources effectively as a project gets larger
- Handling comment spam in blogging environments
- Achieving usability for a technically diverse user base
- The value of an open source community in recruiting people
- The place of blogs in a world of open-ended communication options
Sean Campbell: Matt, could you start us off by introducing yourself?
Matt Mullenweg: Sure. I am a co-founder of WordPress, and I currently lead the development effort. WordPress is open source publishing software that started out as blogging software, but is now starting to be used as a content management system as well.
Sean: There’s been a lot of talk lately about Red Hat’s business model; a lot of people hold them up as the gold standard for a successful model of growth and profitability using open source. There are also rumors that IBM’s going to acquire Sun, and of course, Sun recently acquired MySQL.
Against that backdrop, what can you say about the right way for a corporation to acquire an open source company, especially in terms of getting people to stay with the effort after the acquisition is complete?
Matt: I think that you have to be very cognizant of the existing culture and, ideally, try to change stuff as little as possible. For example, my understanding of MySQL is that 70 percent of people work from home. Trying to change that could be very disruptive.
There’s synthesis with all acquisitions, and it’s important not to break what’s working well. Google kept YouTube with separate offices–basically as a separate operation–because they had already done super well. In general, I think it’s best to change as little possible when you acquire something, particularly if it’s open source.
Sean: Are there areas you think are safer to change than others? It seems that large organizations are particularly likely to want to make changes, for better or worse.
One area where it seems there might be an opportunity there is in the case of a small up-and-coming project that may not have the ideal depth of hardware and back-end resources. How would an acquiring company find the right balance between showering them with unneeded technology, versus dedicating additional, bright people to grow the development side of the project?
In other words, how does the company know how to help, beyond just leaving them more or less alone?
Matt: Depending on how it’s done, making more resources available could be great, although I’d be hesitant about giving them to people directly. Still, having more headcount and infrastructure can spur development.
I don’t know if the guidance here is all that specific to open source, actually, versus general acquisitions, since I think that in the software world, the expectations are the same. In either case, whatever you’re hoping to change or integrate should be one of the first conversations you ever have. That helps minimize the degree to which people feel things like, “Ah, crap. We’ve been acquired, and now I’ve got to figure out their stupid user system.”
If you talk about those issues from the beginning, you avoid making people feel blindsided. There will be places where changes have to happen, but if you say at the outset, “OK, we have 100 million users; you have two. It makes sense to integrate those systems,” then you can avoid unnecessary surprises.
A big part of it is that companies really need to think about how easy it is for people to do things on a day-to-day basis. How easy or hard is it to buy a computer, go to a conference, get a new email account, or bring a new server online? None of those issues really fit into a normal design review process or anything like that, but they’re stuff that people deal with every day.
Little annoyances add up cumulatively, and people’s general job satisfaction can depend completely on whether they have to deal with 1,000 annoyances and huge levels of bureaucracy to do basic, simple things. Particularly if they were very simple before the acquisition and difficult afterward, people will quickly get frustrated.
Sean: When WordPress was growing, how did you scale up the IT expertise you needed alongside the core development skill set? It must have been a complicated transition to go from a strictly developer mindset to managing the large number of servers you have now, and obviously, you had to tackle issues about smart growth, reliability, and uptime, which I assume were not core considerations at the beginning of the project.
Matt: Originally, I and a few of the other early developers were doing all the systems administration, just because we had to. We were lucky in the sense that, a few months into it, we started to get help from a fellow named Barry, who had more of a real systems administration background than we did.
He’s very good about making things super reliable and fast, although he didn’t initially have experience with systems at that large a scale. It was certainly a learning process along the way, but we all agreed from the beginning on commonsense goals like wanting the site to be fast, available, and super reliable, so that was a good foundation.
Actually, to this day, Barry is still our only full time systems guy.
Scott Swigart: Of course, not every open source project is backed by a company that offers hosting as WordPress does. As the company offered hosting of WordPress blogs and that took off and became popular, how much did that drive the development to make sure that this open source project was really being built in such a way that would facilitate it scaling? And how much do you think the decision to host has had to do with the overall success of the project?
Matt: I think hosting could have had a negative impact. We made the decision very early on that the structure of WordPress.com would be basically identical to WordPress.org, which honestly made it more difficult in terms of scaling and a few other things.
At the same time, though, it ensures that recruitment and scalability and everything else that was done to WordPress.com has fed directly into the back-end open source project.
I don’t know how much of an impact hosting has actually had on WordPress’s success, simply because 99.99% of blogs never reach the point where the level of traffic is an issue. On the usability and interface side, it has provided more flexibility, because we are able to test things out in a very time-independent way.
We can do things as a plugin for .org or .com and put it out there and let the market decide whether it is something worth having or not.
Sean: From some of the interviews I have seen, it sounds like you’ve had kind of an evolving relationship with comment spam. Obviously, the problem isn’t going to go away soon. What do you think is the next set of punches and counter-punches in that area?
What do you think the next level is that those folks might try, and looking ahead, how might you try to keep that problem at a low roar or squelch it?
Matt: Luckily, our anti-spam product, Akismet, has remained at a very high level of accuracy. I think the thing that is starting to impact us more is that the spam is being made to look more like human comments. Entries may comment on previous comments, and in some cases, spammers actually pay people to write the entries in.
People don’t realize the extent of this sort of thing. The spammer may leave a compliment on the blog, and the blog owner is flattered. They may notice that the URL looks a little weird, but that seems secondary to them.
Unfortunately, if you are going to keep spam off your blog, you have to be a little bit paranoid and really check all the URLs and everything.
Sean: What’s the next technological piece you are thinking of adding to deal with the spam that really doesn’t look at all like spam, like when you get a comment that says, “I loved your blog,” and the URL is Bank of America without the F? What’s the next logical move in that area?
Matt: We’re not really planning any major technological changes right now, because we’re catching that stuff. The biggest thing needed is not technological, but educational– just informing people, and educating them about what really is spam or not.
Sean: I have to wonder how solvable some of these problems are. Blogging has progressed from more technical people to less technical people, and that’s what you want with technology, right? You want it to become mainstream and pervasive and really a new paradigm.
You want it to be approachable by people who aren’t technical, because they have interesting things to say, too. At the same time, they are inherently not going to understand some of this stuff. So, how do you guys think about that balance?
Matt: We try to treat our users as much as possible like they’re intelligent people, and we don’t try to dumb it down. Our approach is to provide the opportunity for people to enter into it and learn from the system, and they really do.
On one hand, there is the impulse to make everything super simple, to ensure it’s accessible. On the other hand, you have to consider something like World of Warcraft, which is an incredibly complex system that has tens of millions of people paying $10 a month to use it.
Part of the secret there, of course, is that it’s fun, and I think anything that we can make fun will allow people to have a true learning process.
Sean: A related issue is that it doesn’t have to be for everybody–you don’t have to try to make it one size fits all. I’m sure there are people who can’t figure out World of Warcraft, but many of them don’t particularly have any interest in it. That’s OK–it can still be a massive success.
Matt: I think if you try to build for everyone, you are pleasing no one.
Sean: Usability-wise, WordPress has always been great. I don’t claim to know every single widget and option, but I have a reasonably solid understanding of it, and it has never been painful to figure those things out. You couldn’t say that necessarily about a lot of other open source software.
How are you guys structured internally to maintain high usability? Typically, that gets difficult as a project grows, unless somebody’s acting as a benevolent dictator.
Do you have a final launch authority on usability, or do you test it on user groups, or something else?
Matt: One of the most important things is that everyone developing the software uses it every day. That sounds basic, but I think most software actually doesn’t have that. If you’re developing a word processor, you need to write something in your word processor every day, and if you’re developing blogging software, you need to blog every day.
You also have to realize that for some software, that’s really hard. If you’re developing a Sarbanes-Oxley management system, most of your staff are not going to need that at a personal level.
Sean: [laughs] Right. Not all your developers are bankers in their evening jobs, so they’re not going to be able to understand all the ramifications of some UI change.
Matt: Still, in consumer software, that’s typically completely possible, because every developer can also be a beta tester, and I think that’s really important. The output of that is that you get everyone thinking from a user point of view, versus an implementation point of view.
Part of my job is also finding the best people in the world, who can give some really powerful input into what we’re doing. But more, I think you just need to have sort of a set of shared principles that everyone agrees with and everyone can believe in, like a philosophy.
One of our philosophies is to include as few options as possible. We try to make things work by default and not burden the user with lots of options. That translates into everyone thinking about things in a slightly different way, and being able to articulate those things into the minds of every single person adds a bit of nuance that would be lost if we had separate people who did development and others who did design.
Sean: I see companies that build “line of business” apps trying to add more and more design, in fits and starts, to their development process. You’ve got trends like consumerization of IT, and you’ve got a workforce that expects apps to have a certain fidelity. If your users walk in and see some ugly, heinous line of business accounting or HR or travel app that’s been around for 10 years, they’ll say, “You’re going to make me use this?”
In terms of hiring the right people, do you feel that open source, in general, makes it easier to identify candidates that rise above the middle and get to the top? In corporate development, you can say you worked on a project, but it can be difficult to communicate what you did and to prove it and to really show a long stream of communications in solving problems and interacting with a community.
It would seem like that makes the job of hiring easier than if you were trying to draw on a team of closed source candidates for a closed source project.
Matt: I agree.
Scott: What are some examples of how that has panned out for you?
Matt: When you’re doing any sort of hiring, if it’s a person you’ve worked with in the past, you know how they’re going to operate in difficult situations. In reality, it’s more important than anything on a resume to know how they’re going to interact with other people and the product, day to day.
Open source allows anyone to jump into that process, without any gatekeepers, and the pool of people is often broader and deeper. There’ll be people in Uruguay or Brazil or Mexico or some place in Europe, and when it comes time to hire, you’re like, “Well, so and so would be perfect.”
And it’s a complete meritocracy.
Sean: In a commercial environment, there’s often the supposition that if somebody comes from XYZ company, they are used to a certain corporate culture, and that often becomes a consideration in the hiring process.
When an open source team is hiring people from other open source teams, does a similar sort of interplay tend to happen? For example, does it come into play when you’re looking at a given effort, how that project tends to organize its discussion, its voting process, and its way of arbitrating between usability versus technological issues?
Matt: Absolutely. For example, we’ve worked with some people who came from the Mozilla project, and they have a certain way of doing things that involves a little more bureaucracy than most open source projects, just as a result of their size.
A lot of factors like that contribute to someone’s background, although there’s never a direct correlation that tells you that, if someone worked on X, they’ll be perfect (or not) for Y. There’s always a learning process as they’re becoming part of the culture, so there’s no secret sauce there.
Sean: Considering various ways of communicating on a computer, there’s blogging, and Twittering, and so on. It seems that the average user probably has a limited number of ways they want to go about communicating on a given day, and most of them adopt just a few; e-mail and IM are the other obvious examples.
Do you see any ripples in the future, where people might be getting closer to moving away from blogging toward something else? Maybe toward something that’s more rapid fire, or less asynchronous?
I’m not trying to position Twitter or any other specific option in that place, because I see them as more complementary than anything else. At the same time, I also see that with all the different options people have for communication, eventually Joe User will get saturated and be forced to pick just a couple.
Matt: I think that complementariness is foremost right now. Twitter and WordPress aren’t super tied together, but you can imagine that every blog post I do should be posted to Twitter, and probably some of my Twitters should go back inside my blogs.
The reason I personally like blogs is because they serve as the best online profile I could create. More than my Facebook, my Friendster profile, or any of these others, my blog is really a representation of me. Because I have complete control over it, it’s a bit of me. By now, it contains seven or eight years of history of things I was interested in, including photos and all kinds of things.
As I interact with things like Flickr, Twitter, and Facebook, I find that I want to aggregate them into my blog, because that is sort of like my home base. I think that people online will always need a home base like that, although the other things that they interact with will be constantly changing.
I did not see YouTube coming at all, but it’s hard to imagine something that has had a bigger cultural impact in the past couple of years, and it’s another way that people now share, aggregate, and interact with one another.
Sean: Blogs have very rapidly gone from being a new thing that geeks did, to spreading out to non-technical people, to really having an impact on politics, policy, and news. In some ways, people probably view blogs as being kind of mundane.
You don’t run into a blog and say, “Oh, what’s that?” What still surprises you? What have you run into recently around the notion of blogs that is fresh and new, and catches your attention?
Matt: I am always fascinated by things that capture peoples’ hearts and minds. In the past couple of years, Twitter has captured a huge amount of press, and I think a good chunk of peoples’ thoughts, but if you compare it to Facebook or YouTube, it’s a drop in the bucket.
When you can make the interactions with the application or website into kind of a game, that’s when I think people start to really make it part of their everyday life. If you’re a part of the game, you’re also facilitating communication or imparting information, or people are learning or reading, and that can be very rewarding.
Facebook in particular has incorporated game-like elements to the nth degree. It is literally addictive. I have peers–particularly those who are still in college–who spend hours per day on it. How is that possible? It’s ridiculous.
Sean: Right. It’s not a game in the World of Warcraft sense, but it has a similar semantic feel that makes you want to play along. You get drawn into the interface, and even if the interface has a hiccup or two, you just drive around the pothole, because you want to play the game.
Scott: That is what we need to solve the recession–we just need a game-based economy.
Sean: Well, I’m glad we solved the economic crisis. This conversation has been time well spent.
Matt: Thanks. I enjoyed talking to you both.