[Freedombox-discuss] In-the-cloud infrastructure and business involvement (was: distributed DNS)

[ian at churchkey.org]

On 03/15/2011 05:26 PM, Clint Adams wrote:
> If we design for some kind of petname system, I can address
> your FreedomBox as "Bjarni's FreedomBox" or "BRE's site"
> or "That PageKite Guy's Dreamplug" and that name can
> map to something.pagekite.net or
> http://axqzzpkfwezf3kku.onion/ or tcp:ohvei9ab.fb2fb:8080
> or something that uses nothing resembling DNS or IP.
> As addressing schemes change, I want to be able to switch
> between them easily, but still retain the exact same
> human-readable addressing and the exact same FOAF-type
> relations.
> 
> Should we not do that, and say "Here are three different
> service providers you can choose from to sacrifice your
> personal privacy to, but you are free because you have
> choice," I think we are missing the point.

I don't think we're on different sides of this actually. I certainly
hope we will have a system for addressing someone's FreedomBox through
multiple schemes, both internal to the FreedomBox network and publicly
facing for all the communication we want to do with people and
organizations who don't have FreedomBoxes.

It is primarily when trying to talk with people who don't have
FreedomBoxes that I think we will run into the need for things like
Dynamic DNS. Whether the organization running that system is a
non-profit operating to support FreedomBox users worldwide, or some
super-node box my friend runs, most of us are still going to be reliant
on someone with a publicly route-able IP address if we want to talk to
the world at large.

If we're not going to end up in the devil's choice situation you
describe, I think we need to build technical and legal infrastructure to
make choosing people whom we have a /reason/ to trust an easy part of
setting up your FreedomBox. Maybe that means setting up non-profits,
maybe that means additional relationships in the web of trust, or
setting up a bunch of public TOR search nodes in the Pirate Party
offices, I don't know what will work best.

On a side note, I think there are a number of services that are
naturally centralized and that will be quite difficult to re-implement
in a different manner (at least until someone cracks the distributed
search problem). Social networking is the biggest of those, followed
closely by the related social tools like linked-in and online dating.
These systems only work when they are supported by a network effect, and
it is not clear to me how you establish such a network effect without at
least some minimal phone book-like central listing.

Once everyone knows how to contact each other, you can operate as
independent nodes, without the need to keep using the phone book, but
the major reason people join such networks is precisely because they
need to find out how to contact someone, not because they are impressed
by how much better Facebook's messaging client is than gmail. That was
primarily what I was trying to get at with the "Dynamic DNS Facebook"
link (http://churchkey.org/2010/03/17/dynamic-dns-facebook/))

-Ian


All of these systems rely on the network effect of being able to find

Since we're going to be dependent on them,


 if we want people on the general net to be able to communicate with us.

I just think we need a reason to

 If that organization ends up maintaining DNS records

More generally, there are some services that naturally favor centralization