A blog forum to provide deep dive analysis and community conversations about software development models. For more details click here.

OSCON is just around the corner, and in this conversation, we chat with true open-sourcer Edd Dumbill – co-chair of the conference and hacker behind the scenes.

Scott Swigart: To start us off, please take a second to introduce yourself and your involvement with OSCON.

Edd Dumbill: Sure. I have now been the program co-chair of OSCON for three years, and my colleague Allison Randal is the other program co-chair. I took over from the legendary Nat Torkington.

I wear two hats at O’Reilly. In addition to being program co-chair for OSCON, I’m also responsible for the software that runs the conferences, so I kind of get to hack on both sides of the fence.

Before I started working with O’Reilly, I had done a couple of startups and been an open source hacker for a while. I contributed to the GNOME project and was a Debian developer, so I’ve been in and around the open source world for quite a long time.

Scott: How many years has OSCON been going now?

Edd: This is the twelfth year, I believe.

Scott: Obviously, you want to keep it fresh every year, so how do you track what the trends and interesting movements are in open source?

Edd: That’s a great question, and it’s actually the part of the job I enjoy most. We go about it in two ways. The first is we’re lucky to have a great response to our call for participation. This year we had well over 700 proposals.

Considering that we could only accept around 120 of those, we unfortunately had to turn away a lot of really great stuff, but the spectrum of responses paints a pretty good picture of what’s trending and what’s not.

The other thing is that O’Reilly tracks projects and the industry as a whole day in, day out. An example of that effect is the health IT track, which we’re doing at OSCON this year. We identified health technology as an area that’s ripe for the kind of collaboration that comes with open source.

We’re betting on that, and we opened up a separate call for participation for health IT, which we’ve had a great response to. We hope not only to cover trends there, but also to be part of getting new conversations and collaborations going.

In a similar vein, we’ve also set up two summit days in parallel with our tutorials. The first summit is an open source and cloud summit, where we’re gathering a lot of leaders to discuss the place of open source in the cloud, as well as what the future holds for emerging standards and collaboration.

At the hacker level, we’re doing a summit on the Scala language, which is an up and coming JVM language that has both enterprise and hacker credibility. It’s been used in places like Twitter and FourSquare, places where there are a lot of scalability concerns. We’re organizing a half-day Scala tutorial on Monday, and then a full day of Scala-related talks on Tuesday.

To come back to your question, we observe and try to shape some trends. NoSQL databases will be featured strongly this year. There are also things we identify as up and coming, and we try to give them space to grow and provide a forum for our community to get to know each other.

Scott: Five or ten years ago, it felt like open source was trending toward a separate, parallel universe. We would have an open source operating system with an open source web server on it. There would be open source office productivity tools, open source development environments, open source compilers, and it was all very distinct from the proprietary side of the industry.

Looking at OSCON now, there’s stuff about Android and iPhone, and even though Android is open source, it’s not really community developed. There’s also a lot of material at OSCON about cloud infrastructure, and that area involves a lot of proprietary players, whether it’s VMware or Amazon. Amazon built an infrastructure using open source, but they’re not exactly open.

It seems that a lot of proprietary stuff shows up in direct conjunction with open source now. There’s even beginning to be a thriving open source ecosystem around those proprietary offerings, whether it’s iPhone or whether it’s a certain cloud provider or something like that.

You mentioned that your involvement with open source stretches back a ways, into some Debian development and that kind of stuff. How have you seen the perception of open source and proprietary software evolving over that time?

Edd: I think there’s always been a spectrum, and that spectrum still exists today. At one end, you have the free software movement, which wants a complete stack of completely free software. At the other end is corporate open source, where the control is still very much with the organization, but they’re using open source as a tool. And there’s a spectrum that lies in between.

O’Reilly has always been pragmatic, recognizing that while the free software movement is a very important thing, you don’t have to subscribe to it to get value from open source. The key to open source and sharing source, and also really open data and open standards, is that there’s a lot of value to be had without subscribing to either extreme.

Scott: How does that play out with transitions like the current one toward the cloud, which is really causing some new types of convergence between the open source and proprietary worlds?

Edd: One of the really important questions facing open source right now is with software as a service, what is the place of open source? Previously, if my text editor was broken, I could just develop it and hack it, add a new feature and submit the patch. What happens when I can’t do that?

You’re seeing a slow uptake in the popularity of the Affero GPL license. Not only does it require you to publish the source of you make any changes to it, but, but if you were running this as a service, then you must publish your source.

A poster child project for this is the StatusNet project that Identi.ca runs on top of, but we also saw an announcement this week that Sugar CRM v6 will be AGPL. There is a free software reaction to the web services, and it will be interesting to see how that develops.

We also have Stormy Peters giving a keynote on that topic, and she’s also hosting a Birds of a Feather session and a bunch of lightning talks on it. I expect OSCON will be a forum this year where we can talk about our reaction to cloud and open source.

With regard to what you’re saying about how open source has transitioned away from free software idealism and so on, corporations are catching on more and more to the fact that open source is a very useful business tool. Tim O’Reilly made the point that you can see open source as a tool used by the second player in the market.

Taking phones as an example, the iPhone has a lot of dominance with a completely closed platform. The key for Android is that it’s open source, so they’re getting a competitive advantage by opening up. We are seeing that particularly in the mobile segment, but in other places too.

Open source is becoming a business tactic, and OSCON will have a lot of content to help guide businesses.

Scott: We’ve interviewed people from a lot of companies that are wrapped around open source projects. We have found it to be quite common that somebody had an idea, built an open source project, and did the care and feeding of it for a while.

Then, when they got a lot of people using it, they launched a company around it to provide professional services, customization, support, and so on. It may be a bit of a stretch to say that most popular open source projects have some kind of company kind of wrapped around them, contributing code, but sometimes it seems that way.

What are your thoughts on that? Can you give us a little bit of preview about the bullet in the conference description, “use open source effectively as part of your business strategy?”

Edd: In the business track, we’re targeting two kinds of audiences. The first is corporations that are thinking about open sourcing their code bases. OSCON is a great place for people to get oriented into the open source platform.

The other audience is businesses who are using open source. You can leave yourself open to a number of problems if you’re not actively tracking your use of open source and the associated licensing. Those issues can be especially complex in acquisition scenarios.

In general, open source developers need to pay more attention to matters of law. We have a session from Bradley Kuhn and Karen Sandler, which is proving very popular already, called “The Bare Essentials of Legal Issues for Developers.”

The goal of that session is to teach what developers need to know about trademarks, copyright, and patents. People can’t just release open source code and then forget about it. The future of their project is vulnerable on all three of those fronts, and people must not be ignorant about it.

Scott: At this point, it seems to be inevitable for a software company to run into open source, whereas ten years ago, everything they were using might have been proprietary.

To look at the really big examples, IBM obviously interacts with open source and contributes to it a lot. Oracle just acquired Sun, so there’s a ton of open source in their portfolio. Even Microsoft now contributes to the Linux kernel and has a whole bunch of little projects they’ve released as open source.

If your company gets really huge, it will almost certainly interact with projects and contribute code, but more along the lines of what you’re alluding to, your developers are just going to want to use it.

The difference between an Apache license and a GPL license can have huge ramifications if you’re bundling the code with something you’re distributing.

Edd: Absolutely, and companies need to be aware of that. There’s another aspect that we’re also covering. I think business and community in the open source world are very much interlinked. You can’t really use open source in isolation from its communities.

We have a whole community track that discusses things like how to manage a community full of contributors, how to be an effective leader in those places, and how to foster diversity.

We have a fascinating panel that Leslie Hawthorn’s put together about financial incentives in open source. Leslie ran the Google Summer of Code Project, and she just recently left Google. I’m really looking forward to that, just to hear the panel talk about what works and what doesn’t work and how you can effectively use money to accelerate and encourage open source projects.

Scott: It’s a fascinating topic, and a lot of open source projects have failed because a company thought it was going to be easy. They just threw the code out there, but the community didn’t come, or maybe it fractured. There are a lot of very smart people working with open source, and a certain percentage are very sure that what they believe is absolutely right.

Community managers need to address the issues that come along with strong personalities, and they need to incent people to do work that nobody really feels like doing, even though it’s critical. Getting community contributions requires a different mindset than when you have a hundred developers on staff.

There’s also a whole domain of expertise associated with deciding strategic issues like whether to follow the aquarium model or the community model.

Edd: That’s true, and the community as a whole is only now finding out a lot of those things. There’s only a handful of projects that you can look to for a ten-year span of experience, and some of them have gone through pretty radical overhauls.

There’s another presentation from Dave Beckett, who works for Yahoo! now. He’s been maintaining infrastructural projects like the PNG since 1994. His presentation is called “Mature Open Source: Making Your Software Last a Decade.”

What’s fascinating about him is that he’s just one guy. As opposed to the corporate side with incentives and whatnot, he started off more or less as a solo hacker and kept his projects going. As time has gone on, he has slowly expanded the pool of participants and so on.

I’m glad that we’re hearing the story from both ends. The corporate side is talking about taking their projects open or funding projects, and the lone hacker side is talking about how to stick with something and slowly build participation without going crazy.

Scott: OSCON is in a fairly unique position to really understand open source developers, because you’ve been running this conference for 12 years now.

Open source has evolved with corporations getting more involved, but there’s still that core constituency of open source developers. Are you guys ever surprised by what resonates or doesn’t resonate with those developers?

Edd: When you put together a conference, you’ve got to have several things in mind. First and foremost, people are paying to come there, so it’s got to be useful and relevant. Over and above that, it has to educate and challenge as well.

As program chairs, we play our part both in recognizing people in the community who’ve got something different and challenging to say and inviting people who we think will stretch people’s minds and thinking in our keynotes.

Scott: So you’re not shying away from controversy, in other words. If you think somebody has a really intriguing point of view, you’re OK with it being somewhat controversial?

Edd: I think so. For example, we have been sponsored by Microsoft, and they used our conference to announce their open source initiatives. When they first did that, it was the most frosty, hostile reception that they probably feared.

[laughter]

Still, although there was controversy, the OSCON crowd is a civilized and human one. In fact, I’ve never known a warmer bunch of people to work with, and to be among for a week. Nevertheless, because we’re geeks, we debate with a certain degree of ferocity.

To their credit, Microsoft stuck with it, and by year’s end, they started to deliver on a variety of fronts.

Sean Campbell: You obviously have a defined process that involves committees and so on to determine which people are asked to present at OSCON, and I imagine that there are certain best practices to get invited as part of that process.

Without naming names, obviously, are there two or three things that you see a lot of people do that either help or hurt their efforts to appear as speakers or panelists at the conference?

Edd: First, we certainly like to hear from the principals themselves. It’s generally a disadvantage if a PR company submits a proposal on your behalf, for instance. When we deal with the actual speaker, that person has much more of a personal investment, and we have a clearer understanding of what’s going on.

We do sometimes accept things that are submitted by PR, and that’s the way certain companies prefer to operate. The reason that it typically puts them at a disadvantage is that when a geek writes for other geeks, it tends to speak more directly to the audience. If it’s written by somebody else, it doesn’t quite grab you in the same way.

The second thing is that we like to see a fair amount of detail. We have 700-plus proposals, and part of the way we can be persuaded that you know what you’re talking about is that you’re able to write in a coherent manner and you give detail about the subject matter.

Not every superstar hacker is a superstar presenter, so the ability to present your thoughts clearly, even in just a description and abstract, is a pretty key indicator for us. We also look for the subject matter to be fresh and concrete, rather than being based on buzzwords.

For established projects, we tend to get a lot of proposals that describe “the state of X” or “what’s new with X in 2010.” That’s all well and good, but frankly, you can read that information off a blog, and we tend to de-emphasize those.

In the past, we’ve put them in lightning talks, where maybe 16 projects got five minutes each to say what was up and coming, but we’re not even doing that this year. These days everyone can read a blog or a changelog or a tweet or whatever. We’re looking for a lot more practical experience, or somebody to put forward an interesting new spin or viewpoint.

Sean: Someone who’s used to giving monologues or who is just too sensitive may not be a good panelist. For a panel, you may want someone who is a bit of a contrarian to make things interesting, without being so far in that direction that the conversation seems fractured.

How do you balance looking for someone to participate on a panel versus an individual speaking slot?

Edd: I think you said it pretty well, and to be honest, we don’t actually accept a huge number of panels. It’s very difficult to do a good panel, and I think some of it comes down to the personal preferences of the chairs.

Neither Allison nor I like to do too many panels, largely because if they don’t go well, they don’t tend to be very useful means of communicating to people.

Sean: How do people get involved in the actual planning side of the conference? Particularly because O’Reilly is a for-profit corporation, what is the experience like for someone who is not an O’Reilly employee and wants to participate? What channels do you direct that kind of feedback to?

Edd: We have a variety of channels, and we do rely very heavily on community contributions, with the first point of contributing obviously being through the call for proposals. A lot of longstanding friends of the conference contribute in that way.

We try to listen to the largest possible swath of people in and around the community, including our company. We do have a formal program committee made up of 20 to 30 folks active in the various orbits that OSCON covers.

We’re always looking for people to invite onto that committee who are active in their particular part of the open source world or interested in how to make OSCON better.

It’s a broad conference, so Allison and I as chairs need to rely heavily on the expertise of people in various areas. We have a Perl sub-committee, and people in Python, PHP, Postgres, and so on. We do some degree of self organizing, and they have little committees of reviewers who review all the talks.

We try to make sure that every single proposal is seen by at least two people, and very often a lot more than that. Everything is read and considered and voted on.

Scott: To call out a technical point, I see that you’ve included NoSQL as a topic in cloud. How intersected do you feel like those two are? In other words, would we have a big interest in NoSQL if it wasn’t for the cloud?

Edd: That’s an interesting point. We have three broad topic areas that we see as the growth areas this year: NoSQL, cloud, and mobile. All of them are interrelated, and to understand that interrelation, you need to dial back a few years, when we were talking about the LAMP stack.

Then we were talking about web applications, and the web sews all these topics together. The question of whether we would be as interested in NoSQL if we didn’t have the cloud is interesting from the standpoint that they both stem from the massively distributed and popular information network that we all use, called the web.

Perhaps the aspect that brought NoSQL to the biggest audience is the need to scale web applications and to be able to build inverted indices from regular SQL databases for performance reasons.

The vast majority of NoSQL applications sit side-by-side with an SQL database. There’s a collaboration and a cooperation going on there, and it’s a hybrid architecture.

Another interesting spin to NoSQL concerns graph databases. They have been around for a good while, but the rise of social networking on the web has finally given us a graph a lot of people care about. Again, you’ve got web applications driving that side of it, too.

Mobile is now quite interlinked with everything else. It is no longer an option, whether to add a mobile interface to your application. It’s kind of expected, rather than a value-add, if you’re launching any kind of commercial service.

Looking at the overall package as a bunch of thin clients and a bunch of chunky servers at the back end, it’s clear that mobile and cloud go hand in hand. If the client can do less, then the server has to do more.

All these issues are growing together, and the other very popular thin client, of course, is the web browser. One must not underestimate the importance of the HTML5 feature set. You’ve basically got feature-equivalent thin clients in the mobile world and the actual desktop.

One of the bets we’re making this year is that if you’re looking to target everywhere, including Android, iPhone, and the rest, then web technologies are the sane choice, as well as the most open route.

A lot of people are rightly worried about developing solely for an Apple platform, when they can change the terms and conditions, and blow what you’re doing out of the water.

Scott: In another conversation earlier today, someone observed that people get too hung up about building iPhone applications with their proprietary SDK, Objective C and that kind of stuff. Most of those applications could be built as web applications, and they can even run locally on the phone, with local storage.

There seems to be an awareness problem, where people aren’t really considering that option. Do you think that web-based approach is going to catch on?

Edd: I think this is the year when people are going to figure it out, and we’re actually working on that internally at O’Reilly at the moment. We develop a bunch of apps, and a lot of our books have been successful in that format on the App Store.

We think we need to provide more mobile interfaces to the things we do, and personally, I think the conference needs to be one of those.

Developers can either put up with a lot of risk and delay from the App Store, or they can do a perfectly great app using just the mobile browser. The poster child for this is Google’s mail, Gmail, which delivers a great experience on the iPhone.

Scott: PhoneGap makes a web app to run local on Android, iPhone, and all that kind of stuff. It looks just like an app, but the whole thing’s built using HTML5.

Edd: We have folks presenting at OSCON on that, and I mentioned several of the cross-platform talks about that sort of thing.

Scott: You mentioned that this is going to be the year when this model takes off. Since O’Reilly has the ability to put a spotlight on things, talk a little about the notion that, “With great power comes great responsibility.”

You have observed that using proprietary SDKs and so on can require developers to start over from scratch in order to go from iPhone to Android, with very little ability for code reuse. How concerned do you think you need to be with regard to companies’ terms of service and so on, when you advocate for a certain development approach?

Edd: Honestly, a lot of those things come down to personal values. The broad picture is that we’re obviously very much in favor of openness, transparency, and so on, but we’re not breaking with vendors consistently.

We will stand up and make a big deal of it when it matters, like Tim O’Reilly’s famous spat with Jeff Bezos’ one-click patent, for instance. We also stood up and spoke out directly in favor of Microsoft when they were being hit with a silly patent dispute concerning their web browser.

I don’t know that the conflict between various mobile platforms is necessarily any different from Windows versus Linux versus Mac, in the sense that all these different platforms have different terms of engagement, and open source plays a role in each.

As far as mobile’s concerned, we’re here to help equip developers, so we have a great range of books on iPhone, Android, web standards in HTML5, and so on. Our role is really one of educating, rather than taking a stand on how people should develop.

Scott: I want to be sensitive to the time. Any final thoughts about OSCON that you’d like people to be aware of?

Edd: I’d encourage everyone who reads this to look through the OSCON program. There’s a wealth of topics that we’ve not been able to touch on here. For instance, we have a fantastic hardware track, covering Arduino, things like the Beagle Board, and plug computing apps.

It’s become very easy to prototype, and there are a lot of fun things we can put out in the world to process data, connect to the cloud, and monitor and augment reality. That’s a fascinating area for us.

We have an operations track,, which is very much following the rise of developer operations as a hybrid discipline, and something else the cloud brings with it.

You can’t really run software for the cloud without taking its architecture into account. When you have ten thousand instances of a machine, you need to allow for that, so the Dev Ops discipline is very much growing.

There’s so much there, I could go on forever.

Scott: It’s obviously a great conference. These are technical people, but at the same time, OSCON really tracks trends, conversations, and debates. Having Microsoft come in was a controversial thing to do, but you guys felt that it was important. It seems that you really live by a philosophy of not shying away from controversial topics.

Edd: That’s right. Community means a great deal to us, and we regard it as an enormous privilege to have this conference. The attendees and speakers each year are what makes it a success, and it’s absurdly gratifying to stand on stage and welcome people.

It’s a fantastic collection of people who are inspiring, entertaining, and controversial. I really couldn’t think of a better job.

Scott: Thanks for taking the time to chat.

Edd: Thank you.

Comments (0) Posted by scottswigart on Monday, May 31st, 2010


You can follow any responses to this entry through the magic of "RSS 2.0" and leave a trackback from your own site.

Post A Comment