Bdale Garbee is well-known in the free software world for a number of
different things: his work with Debian (including a term as project
leader), his work as HP's open source and Linux chief technologist,
membership on several boards (the Linux Foundation among them), and a lot more.
He's also known for giving talks at various conferences about another
passion of his, model rocketry, and specifically how open hardware and
software can be used to control and track those rockets. So when he said
that his LinuxCon talk
was a rare example of a "talk I would rather give than a rocket
talk", it's a pretty good indicator of how important he thinks the
topic, FreedomBox, is.
The FreedomBox project is an effort
to create personal servers that will run on cheap, "plug computer"
hardware. While the software will be designed to run on hardware installed
in the home or elsewhere, the focus is on in-home use. In some
jurisdictions, Garbee said, there is a big difference between how data
stored on a computer in the home vs. one elsewhere is treated in a legal sense.
The project also wants to "contribute to privacy-respecting
alternatives to social networking". In today's world, people are
uploading personal data to services like Facebook without any real
guarantees that the data will still be there in the future, and that they will
always have access to it. In addition, the terms of service can change
over time, as do the privacy settings and policies. Garbee was careful to
point that the project (and the FreedomBox Foundation) would
not necessarily be creating these social networking alternatives, but would
be collaborating with those who are.
Another important part of the FreedomBox idea is to support mesh
networking. As we have seen in the news recently, activists and political
various places are too dependent on centralized services, especially
communications services. We already have the technology to build mesh
networks that could be used to route around repressive governments, or just
repressive ISPs, he said. If two neighbors have different ISPs, with
different filtering policies, a mesh network between them could potentially
avoid those problems.
Debian and FreedomBox
There is a "high correlation" between the goals of the Debian
distribution and those of the FreedomBox, Garbee said. There is also
"no better place to find a strong technical infrastructure"
than in Debian. In something of an aside, he also noted that while Linux was celebrating its 20th
anniversary at the conference, Debian is celebrating its 18th
anniversary, which is truly "mind-boggling", he said. There
is no Debian company or corporation, it is made up of individual volunteers.
It also runs on all of the relevant architectures. All of these things explain
why the FreedomBox software is Debian-based.
In addition to all of that, there is a fair amount of truth to the
statement that "all free software gets packaged for Debian",
which gives the project a good base. It can use the same bug tracker and
build environment that Debian uses as well. Many of the pieces that are
needed for FreedomBox are already packaged or being worked on within the
But FreedomBox does not plan to be a Debian derivative, and will instead do
all of its work within the distribution. One of the goals is that every
stable release of Debian will have "everything needed to create
FreedomBoxes", Garbee said. So users can either buy a plug computer
and install FreedomBox themselves, buy an off-the-shelf plug computer with
the software pre-installed, or find a cast-off computer and install it
there. One of the big advantages of that approach, he said, is that no
matter how successful the FreedomBox project ends up being, all of the work
and code will always be available in Debian.
The FreedomBox Foundation (FBF) was founded by Eben Moglen, who has "done a
great job articulating the need" for such a device. Moglen asked
Garbee to join the board of the foundation in order to establish and chair
a technical advisory committee (TAC). The TAC exists "to make the
board understand what the technical issues are", he said, and it is
not a "top-down design group". That work will be done in the
soon-to-be-established working groups.
The FBF is not a large organization with "a lot of resources and an
army of coders", Garbee said. The technology is not really the hard
part, he said, at least for most of the people in the room. The much harder
part will be the user experience because the FreedomBox has a "much
broader audience than just those who are building it". If those
others can't understand how to use it, "we will have failed".
So far, that's an area where, unfortunately, not a lot of work has been
done yet, he said.
There are other tasks that the FBF is taking on, such as fund-raising, outreach, and publicity. Those things are important and are a persistent problem for any
non-profit organization, he said. Another non-obvious thing that the FBF
can do is "industry relations". At some point, hardware vendors should be
willing to build and ship products with FreedomBox pre-installed. That may
require NDAs, which is not something that most free software developers
want to deal with.
The TAC has been formed with Garbee as the chair. Five others are on the
committee as well: Jacob Appelbaum, who is security researcher and core
member of the Tor project; Sam Hartman, a Debian developer and security
consultant; Sascha Meinrath, author and mesh networking researcher; Rob
Savoye, GNU toolchain hacker and embedded systems developer; and Matt
Zimmerman, who is a Debian developer and former CTO at Canonical.
Over the coming weeks, Garbee said, various working groups will be
established to work on the disparate pieces that make up FreedomBox. There
are a lot of different conversations going on in the mailing list, and they
are often getting derailed by people who are focusing on a different piece
of the problem. These working groups will likely be "instantiated as
separate mailing lists" and will be tasked with a specific piece of
the problem. The output may be code, packages, or recipes, he said.
Garbee is "looking forward to getting them going".
DreamPlug reference platform
has been chosen as the initial reference platform for FreedomBox. Part of
the requirements for the FBF's Kickstarter fundraising campaign was to
deliver hardware to some donors, and the DreamPlug will fill that role.
While the hardware is reasonable overall, he said, there are still some
frustrating things from a free software perspective. Marvell created most
of the hardware inside the DreamPlug, and has generally worked well with
the community, but there were still some driver and source availability
problems. Most of those have been resolved except for a firmware blob that
is required to run the Marvell wireless uAP device.
The idea behind the choice of the DreamPlug is to pick a specific target,
hardware is fairly capable. It has a 1.2 GHz ARM processor, with 512M of
RAM, 2M flash for u-boot, and 2G of flash for filesystems. There
are also lots of IO ports, including two gigabit Ethernet interfaces, two
USB 2.0 ports, an eSATA 2.0 port, an SD socket, and more. It also has audio inputs
which didn't seem useful at first, he said, until someone pointed
out that they
could be used for random number generation.
One of the areas that has been extensively discussed within the project is
the idea of "establishing trust". OpenPGP keys are "about
as good as it gets" in terms of storing public/private
key pairs, he said, but the trust relationship problem still isn't solved.
Noting that the target audience may be more likely to have smartphones, the
project is narrowing in on solutions that would allow an initial key exchange
using the display and cameras of smartphones. A phone app could gather
these keys up when people meet face-to-face and then allow them to be
installed on the FreedomBox.
In addition, lots of work on the FreedomBox went on at the hackfest that
in Banja Luka, Bosnia and Herzegovina at the end of July.
The focus was on assembling an initial development image for the DreamPlug
and identifying and integrating an application into that image. While lots
of progress was made, and an application was identified (an XMPP-based
secure chat client), they didn't quite get there during the hackfest.
There were also
several FreedomBox talks at the conference itself and Garbee recommended viewing
of those talks.
Going forward, he said the team is "single-digit days" from
releasing initial development images for both the DreamPlug and for x86
virtualization for those who don't have the hardware. There is ongoing work
to use Monkeysphere for
identity management with OpenPGP keys. Work on selecting and integrating
specific applications that deliver "functionality implied by our
vision" is underway, starting with the secure XMPP-based chat
stack. The plan is to do periodic releases until "we achieve
1.0", Garbee said, but he won't say when that will happen,
There are a number of ways for interested folks to get involved, starting
with being "conscious about privacy and other freedoms in all that
you do", he said. Experimenting with the software and helping to
refine the list
of alternatives to the proprietary cloud services would be
helpful. Joining a working group or helping to select Debian packages (and
determine the right configuration for them) are additional ways to help.
Of course, financial contributions to the FBF are always welcome.
In answer to audience questions, Garbee reiterated that Debian was chosen
for pragmatic reasons and there is no reason that others couldn't put the
FreedomBox stack on top of other distributions. He did not want the FBF to
have to set up distribution infrastructure or be saddled with long-term
security updates, and basing on Debian avoided that. He also said that
off-the-shelf FreedomBoxes are "at least a year away", and it
could be longer than that.
[ I would like to thank the Linux Foundation for assistance with travel
costs for LinuxCon. ]
to post comments)