Fedora leader: ‘We want users to control their own computing destiny’

This is part 2 of my conversation with the Fedora Project Leader, Matthew Miller. If you haven’t read part 1 yet, I recommend starting there first.

Bryan Luduke: There are a lot of Linux distros out there with big, lofty goals. Expand into phones and tablets and television; Take over the world and be the king of desktop computer market-share. That sort of thing. If you could summarize the long term goal of Fedora… what would it be? How do you, as the Fedora Project, measure your own success?

Matthew Miller: “We do have very lofty goals, but they’re not necessarily tied to having Fedora directly installed everywhere or having the Fedora brand plastered on everything. We want a world where free and open culture is widespread, we want people to work together as the default, and we want users to control their own computing destiny — that’s the long-term vision. It’s bigger than just Fedora as king of something.

With our basic charter as a leading-edge distribution, Fedora isn’t right for every situation. If you need a support lifetime of over a decade, Red Hat makes an amazing enterprise distribution which happens to be fed from Fedora work. Every time someone installs RHEL — or CentOS — that’s Fedora success. And, it’s not just the direct derivatives; whenever Fedora pioneers technology that helps Linux and open source advance overall, that’s a win.

Of course, we do want the operating system we produce to be directly successful as well. It’s not as much fun to make something no one really actually uses, and it can’t be a meaningful incubator for new technologies without real users. We can’t just be someone else’s perpetual beta, so Fedora success definitely includes attracting users, too.

For a long time, our basic approach was to make something we felt was awesome, and offer it with a sort of “here, we made a great operating system — here’s all the stuff in it, and we hope you like it”. And, to be bluntly honest, that wasn’t working very well. We were making better and better technology, but our user numbers were on a slow decline.

So, we’ve taken a renewed interest in marketing as a real discipline — and it’s really paying off, as we’re seeing our metrics for Fedora operating system use take off. To measure that use, we have a number of indirect tools. Fedora doesn’t do any direct or invasive tracking, so there’s no way of getting an exact count. It’s more like those wildlife surveys where you do some sampling and try to figure out from there. One thing we track is daily unique connections to our system which serves out security and bugfix updates, and from that, we can see a steep upward curve since we started the new http://getfedora.org, with different Fedora Workstation, Fedora Server, and Fedora Cloud editions.

Obviously, we want that to continue. And, we’re working on new initiatives to measure community involvement. That’s just starting out, but I think it’s a really important way to look at our success. Last year, we had over 2,000 individuals contribute to some of the easy-to-measure parts of the project, like wiki edits, software package updates, and testing of those updates.

+ MORE FEDORA: Understanding the Fedora Next initiative through LEGO (kind of) +

I’m planning to expand that to other areas of the project too — translations, user support and help, design work, and so on. I can go into pretty geeky details on these numbers, if you like, but the overall idea is that having a healthy, growing community is a way we think about our success, too.

Oh, and about phones and tablets — simply put, I don’t think it’s an area where we can win. It’s a really hard area to even break into, and you need both the engineering and big deals with big hardware vendors, and for phones, all sorts of regulatory stuff, and having done it once, you have to then do it all over again. It just doesn’t seem like the best use of our limited resources. I do wish everyone working on this a lot of luck, though — I certainly do want to see free and open source software win on mobile as well as the desktop and server room.

What is it about Fedora that makes it ‘leading edge’? Is that anything like the ‘bleeding edge’ that people have with some rolling release distros?

By ‘leading edge’, I mean that we are part of innovation in open source and operating system software, not just integrating already proven solutions. On the desktop side, you can see that in our investment in Wayland, which is a next-generation graphics stack designed to replace the X11 display protocol traditional with Linux systems. On the server and cloud side, we’re exploring how all the buzz around containers will actually change what users will expect from an OS in the future. For example, check out Project Atomic and Fedora Atomic Host, which is specifically optimized for containerized applications.

On a much more mundane but practical level, we try to be leading simply by having a fast refresh cycle, getting updated versions to users quickly. For example, we tend to have new upstream Linux kernel releases out to users within a matter of weeks, and always follow the latest. With some software, like the kernel, this is usually all positive, but for other software where the developers might not take as much care with compatibility, it can be a little rough. That’s where the risk of “bleeding” comes in, and because we actually want to avoid getting our users covered in nicks and cuts, we try to keep big changes for major releases, so we can communicate in advance about what to expect.

A rolling release model isn’t really about being leading or fast. It just means that there are no major releases or release dates — just a single supported version and an ongoing series of updates. This actually lends itself better to slower distributions that aren’t experimenting or quickly adopting new things, because if there is big change under the rolling release model, users don’t get to schedule it for themselves. They get what’s there when the developers decide it’s ready. If that happens all the time, no wonder it feels like bleeding.

So, rather than doing that, we’re focusing on making updates between supported Fedora releases as painless as possible. With Fedora 24 Workstation, for example, whole-release updates are going to be built into the Software application, right next to the current security updates and bugfixes. You’ll get a notification that it’s available, but can take the next half year or so to decide if you’re ready to take it, and you’ll still get those security fixes for the older release in the meantime.

I think fans of rolling releases such as Arch and openSUSE Tumbleweed might disagree with the whole ‘rolling release model isn’t really about being leading or fast’ thing. 🙂

Oh, I don’t mean anything bad by it. Those distros certainly do interesting, innovative things. With Arch, users manage the rolling release model by closely following the development community. Checking the documentation before updating is an expected part of the process — or, at least, checking after the fact if there are problems. That’s actually part of the fun of Arch, and also part of the secret of their fanatically-great documentation. And openSUSE Tumbleweed has a really slick system for automated testing — some of which we’re now using in Fedora, too, so I definitely appreciate openSUSE’s leadership in that area. But at the same time, the sheer amount of change is much greater from week to week in Fedora’s development tree than it is in Tumbleweed.

For an example of a conservative rolling-release model, look at CentOS. Obviously, the main versions (5, 6, 7) are numbered releases, but each one lasts for many years, and within each one, there’s just one stream of continuous updates. This is actually one of the big differences between CentOS and Red Hat Enterprise Linux — RHEL has distinct point releases (for example, 6.8 just hit beta), and customers can choose to take their time upgrading. When CentOS rebuilds the RHEL sources, they combine everything, and users just get the combined updates — a rolling release. Because Red Hat is so careful about compatibility, that means that this generally just works out fine. And, if you don’t want to take that risk, or want to manage it more carefully, you can pay for Extended Update Support for RHEL from Red Hat.

Let’s end on a bit of a different question. If the Fedora council passed a resolution, tomorrow, that forbade you (and only you) from ever using Fedora again… what Linux distribution (or other operating system) would you choose use?

Wow, that’s harsh! I’d probably hop around a little bit and see where I’d find a community I feel like I can fit into and make a difference in. Maybe something small and unknown, or maybe start a new project from scratch — not because the world needs that, but because it’d be an interesting switch from something as big and with so much history as Fedora.

I’m glad that’s an imaginary scenario. There’s so much going on with Fedora these days, with a large vibrant community building cool stuff. I’m excited to be a part of it for a long time to come.

You have dodged that question most artfully, my friend. 🙂 As you look out at the entire Free Software and Linux world, what is the one thing that you hope to see happen over the next year? One thing. Just one.

Oh, that’s hard too. OK, if I’m choosing just one… there was a story recently where researchers showed that patches submitted on GitHub by women using whose gender is easy to guess from name or avatar image are significantly less likely to be accepted than those from men — but when women use gender-neutral nicknames and images, their patches on average are *more* likely to be taken. This is symptomatic of a serious imbalance, and it’s particularly disturbing that the situation is significantly worse in open source and free software than it is behind the doors of proprietary software companies. So, I’d like to see that get better. I have five or six others, but since you ask for just one, I guess you’ll never hear them. Sorry!”

If someone wants to get involved in the Fedora Project… what’s the best way for them to do that?

It’s easy — go to http://whatcanidoforfedora.org and click the buttons until you find something you find interesting, and then follow the prompts. If that doesn’t work, find me on Twitter (@mattdm), G+, Freenode IRC, or wherever, and I’ll help you get hooked up.