[Freedombox-discuss] We need mesh networking later. Now we need one simple compelling feature users love.

Dave Crossland dave at lab6.com
Sun Feb 20 10:03:58 UTC 2011


Hi,

tl;dr We need mesh networking later. Right now we need something that
does one thing well, that political activists can install simply on an
old device they already bought (PC or Linux-based router) and that
they will love and tell their friends about, because exponential usage
is key.

User Experience Design is the profession that knows how to pick the
one thing, how to make it appear simple and lovable, and how to pass
through the phases of exponential growth. I dearly hope the FBF can
find and recruit a UX Lead of the same calibre and position in the
project as the Technical Lead.

On 19 February 2011 12:59, Luca Dionisi <luca.dionisi at gmail.com> wrote:
> On Sat, Feb 19, 2011 at 1:09 PM, Jonathan Roberts
> <jonrob.one at googlemail.com> wrote:
>>
>>> <quote>
>>> More importantly perhaps is the fact that the ISPs are actively
>>> blocking any attempts to bypass this filtering, with residents in
>>> Libya unable to even use the standard filtering bypass tools. We've
>>> been told that neither OperaTor, HideMyAss, nor HotSpotShield are
>>> working. Even standard PPTP VPN connections out of the country are
>>> being blocked.
>>> </quote>
>>
>> OLPCs have mesh capabilities.
>
> Their mesh capabilities are those of standard 802.11s. They use AODV.
> For what I know they should have a big limitation in the number of
> nodes they can scale up to.
> As well as other open issues (the protocol being reactive, the IP has
> to be chosen by the user and has to be unique, etc)
> Not very reliable for a big mesh.

OLPC and OpenMoko are somewhat similar hardware-software projects to
this, also originating from the software freedom movement with
comparatively little seed funding. They contrast with Android,
ReadyNAS, many routers, set top boxes and other hardware-software
projects that originate from capitalists seeking returns on
investments and which happen to use open source strategies to varying
extents while benefiting from proprietary software's features and big
money.

Those products generally focused on a single use-case while those
projects were diffused in many areas. I suggest that there is a lesson
there.

I also suggest that one of OpenMoko and OLPC's subtle but fundamental
problems from their outset was that their software vision was of the
http://en.wikipedia.org/wiki/Waterfall_model and not iterative; their
v1 was far more sophisticated than it needed to be. This overstretched
the projects and weakened them. In both cases I wonder if this was
spill over from the hardware being necessarily so, and having the same
people in charge of both.

So I would like to advocate focusing on one aspect of the full Freedom
Box vision for at least a year, and picking an aspect that already
runs and works on Android phones, ReadyNAS servers and Netgear/Linksys
routers but just isn't easy to use yet - and making it easy.

https://www.kickstarter.com/projects/721744279/push-the-freedombox-foundation-from-0-to-60-in-30
lays out that in 7 month's time, we will have produced a free software
operating system with at least one application that runs on at least
one particular piece of hardware.

So while mesh networking is tantalising, I suggest saying "yes but
later" to it.

All the ideas being brainstormed here could do well to be logged on a
wiki table and ranked in terms of impact and implementation
complexity. That is certainly the work I hope that the technical lead
soon to be announced by the FBF will do and do well. The choice of
concept to prove with running code in v1 seems singular in its
importance to me.

I'm pretty sure a UX lead would do that work, and I dearly hope the
FBF can find and recruit a UX lead of the same calibre and position in
the project as the technical lead.

Store-and-forward networking was mentioned, and that seems to me to
deliver much of the impact of mesh networking and have much more
feasibility; but even that seems to be over-engineering things.

If OperaTor, HideMyAss, HotSpotShield and even standard PPTP VPN
connections are broken by the state, will a SOCKS proxy be enough? And
would an SSH connection to a random and hardly used foreign IP address
on port 80 still get through? I guess so.

In that case, a webapp that can note the user agent's IP, stop the
httpd and start a sshd with SOCKS proxying enabled on port 80 for one
connection from the noted IP, allows people who can ssh into that
freedom box to run

    laptop$ ssh -D 8080 -Nf my.friends.freedom.box -p 80

and be on a secure line to the world. People have real life social
networks that can transmit the login details needed to use such a
simple v1 freedom box app; store-and-forwarding lists of servers could
be nice, but a fully technologicized solution isn't always necessary
for a techno-social system to function.

"Freedom Box is the name we give to a personal server running a free
software operating system, with free applications designed to create
and preserve personal privacy."
- https://www.freedomboxfoundation.org

Anyone already with a computer that they can leave turned on and
connected to the net 24/7 is a potential Freedom Box owner already. In
a couple of years the easiest way to manifest that potential is surely
to to buy $50 wallwart server hardware and plug it in pre-installed.
But while those devices are becoming available at that price, millions
of people have already bought routers (and Android devices, ReadyNAS
servers, set top boxes, etc etc) on which they can install an OS on
that would turn them into a very simple Freedom Boxes.

Such a simple way to operate and access sshd SOCKS proxies is high
impact, high feasibility - one page of PHP, I expect - and would be
just one package away for any system running httpd and sshd already.
Its even possible to patch something like that into the out-of-the-box
experience of most of
http://en.wikipedia.org/wiki/List_of_router_or_firewall_distributions
- any machine running httpd and sshd which the owner has root on will
do.

Installing OS software yourself feels risky and complicated, and
operating it isn't as simple as it could be, but if we can reach the
public with the right information - how to install new firmware for
their router easily and risk-free and how to use its one single killer
feature - and they have a good experience running that very limited
Freedom Box, they'll want to tell their friends, and get a better one.
Exponential usage growth is key.

I suggest such exponential growth takes time, takes appealing to early
adopters with something simple but compelling, and takes iterative
phase changes. User experience researchers are trained professionals
in finding out what appeals to people, and UX designers know how to
translate that research into something software engineers can use.

"Humanity's greatest danger is our inability to understand the
exponential function."
- Albert Bartlett

Cheers
Dave



More information about the Freedombox-discuss mailing list