[Freedombox-discuss] don't write code - user-friendly configuration

[bertagaz]

Hi,

On Mon, Sep 06, 2010 at 01:12:17AM +1000, Anthony Towns wrote:
> On Sun, Sep 5, 2010 at 23:59, Matthew Johnson <mjj29 at debian.org> wrote:
> >  - The user buys the box and plugs it in. It does discovery to work out whether
> >  there's another FB on the network, whether there's an existing router, etc.
> 
> >  - The user then, for example, browses to a well know name that it's set itself
> >  up as in your local DNS or if it's also your ADSL router, any website and it
> >  transparently redirects you.
> 
> Having (a) DHCP+a captive portal if there's not currently a DHCP
> server on the network, or (b) an automatic dyndns-style registration
> using its serial number (sn3141592.freebox.org) if it's on the
> internet, or (c) a special program that uses a custom UDP protocol to
> find the freedom box on the local network would probably work. Relying
> on an existing DHCP server to have an updateable DNS server locally
> seems a lot to ask.
> 
> Using the serial number to register on a distributed hash might work too.
> 
> (The benefit of the serial number is that (hopefully) the software can
> work it out without asking the user any questions, and the user can
> work it out by just looking at the label on the thingy they bought.
> The benefit of captive portals is it just works.)

The downside of a serial number is that it might help to track you with
the credit card number you used to buy the box. Not sure it is a good
option for a box bringing back privacy (cause I understood it was one of
the key point of this project).
 
> >  - The website reached asks you some simple questions like:
> 
> Wouldn't it be better if it just "worked" straight away? Your toaster
> doesn't ask you how you like your toast -- you plug it in, put the
> toast in, press the button, and then fiddle later if it ended up
> burnt...

Mine (and most of the ones I've seen) does have a button to configure the
way it burns your toast :] To be honest, having a box that just
auto-configures itself in a secure and privacy aware way would be great,
but I'm not sure it's feasable correctly without a minimum of user
interaction.  First thing I can think about is pasword prompt for
example... Maybe it's rather more just a way to find good and
understandable phrases to ask the questions to the average users.

One thing that is important when thinking about building a "better" and
secure network is users learning how it works, or at least pros and cons
of choices they will have to make when choosing how to communicate.
Devices that does all the configuration "transparently" usually don't help
a lot in that, and I think there's more benefits to have a good translated
documentation rather than hiding everything to users.

And finally, I don't understand how freedombox could fight the "evil
cloud" and also propose to use it. Maybe freedombox could have a plugin
architecture so that people willing to use such services could write their
own, but I don't think this project should put any effort in providing the
software to use them.

bert.