Re: [GNUnet-developers] [gnu-prog-discuss] MediaGoblin, now an official GNU project :)

[Luke Kenneth Casson Leighton]

On Sat, Aug 6, 2011 at 10:08 PM, Christopher Allan Webber
<address@hidden> wrote:
> Hiya all,
>
> I'm happy to say that MediaGoblin is now *officially* *GNU*
> MediaGoblin.  I'm very excited about this!
>
> Some information about MediaGoblin if you aren't already familiar:
>  - We're attempting to build a distributed, modern media publishing tool
>   for the web (images for now, but the infrastructure is being designed
>   to also support video and other media types)
>  - We're python based
>  - About that distributed thing: we're currently only distributed in the
>   sense that anyone can run an instance, but the immediate plan is that
>   within the next couple of months we'll begin working on federation
>   via OStatus

 http://ostatus.org/

 fascinating.

 christopher, you're aware that the freedomboxproject (which isn't
about providing people with actual "boxes" at all, it's about bringing
together the software that can _be_ installed on a "box") has been
looking for this kind of stuff, in order to allow people to transition
off of the present non-free services such as flikr, facebook etc.?

 also, out of interest, have you seen this?
http://blog.bittorrent.com/2011/06/30/uchat-we-just-need-each-other/

 btw i can't tell if ostatus has built-in firewall-busting (like the
gnunet infrastructure does).  one of the key reasons why all of these
"federation" projects (e.g. sipwitch) are technically unsuccessful is
because they don't have proper firewall-busting built-in.

the reason why they don't have built-in firewall-busting is because
it's f*****g hard to get right, and takes years to perfect and cover
all the edge-cases.  such as what happens if you have 3 levels of NAT
(including one within an ISP), how do you even _find_ that that's
occurring, let alone cope with it (and no, STUN, TUNSS and UPnP aren't
good enough... on their own)

 many ISPs have designed their infrastructure based around the "you're
dumb, you'll only wanna download and that'll be HTTP boyo: Like It And
Lump It" utterly shit paradigm, such that if there are two people on
the same ISP's local NAT'd segment, it's practically impossible to
open a direct connection between the two, even though it would be
faster and would save the ISP a lot of bandwidth and money.

 gnunet is the only free software infrastructure that we have that has
been designed - somewhat accidentally - to deal with this.  it
contains NAT traversal as well as UPnP, _and_, critically, contains
"forwarding" for when a direct connection (which is undesirable in any
case) all goes wrong.

 gnunet was designed to provide a level of anonymity by "hopping"
packets between systems (in the exact same way that TOR does).  it
turns out that this hopping is crucial to any service that wants
reliable, easy-to-use, zero-configuration-needed non-server-centric
peer-to-peer connectivity.

 personally i believe that the easiest way to achieve that is to get
gnunet-vpn up-and-running (preferably the ipv6 version), at which
point it will be possible to just have a totally transparent network
that will "Just Work".  at a later point, integration with gnunet's
modular architecture would provide some level of optimisation, and
provide anonymity that it is too easy to accidentally ignore (if just
using gnunet-vpn).

 on top of gnunet-vpn, chris, the service that you've written would be
absolutely fantastic.  i'd be interested to hear peoples' assessment
of what the benefits of the combination of ostatus, mediagoblin and
gnunet-vpn would bring.

 l.

 p.s. ostatus specification here:
http://ostatus.org/sites/default/files/ostatus-1.0-draft-2-specification.html